Magnesium

Magnesium is one of the body's most abundant cations, and biologically it is only ever used as Mg2+. It is essential to the structure of bone and teeth, acts as a cofactor for more than 300 enzymes — including binding ATP for every kinase reaction — and changes the permeability of excitable membranes and neuromuscular transmission [10].

One fact should frame everything below. More than 99% of the body's magnesium is intracellular, so a serum magnesium level is a poor predictor of tissue content and availability. There is no simple, widely available way to assess magnesium status in people, and the effects of deficiency mostly go undetected [10]. That is why the trials here keep finding the largest effects in people with low magnesium while having no practical way to identify them in advance.

Magnesium is absorbed through both passive and active pathways in the gut. How much gets in depends on the dose, the food it is taken with, and the body's current magnesium status [8].

Blood pressure has the most consistent evidence. A 2016 meta-analysis of 34 double-blind trials found that a median 368 mg a day for three months lowered systolic pressure by 2.0 mmHg and diastolic by 1.8 mmHg [11]. A 2025 meta-analysis of 38 trials found larger falls in people with hypertension who were already on treatment, and in those with low magnesium. It found no significant effect in people with normal blood pressure [1].

Insulin resistance: magnesium improved HOMA-IR, and four months or more of use also improved fasting glucose [2]. Migraine: oral magnesium reduced the frequency and intensity of attacks across 10 prevention trials [3].

Sleep and mood evidence is thinner. In older adults with insomnia, three small trials found people fell asleep about 17 minutes faster, but the evidence was rated low to very low quality [4]. One of those trials also found longer sleep and better sleep efficiency [12]. In mild-to-moderate depression, an open-label crossover trial of magnesium chloride found a 6-point improvement on the PHQ-9 depression scale within six weeks [13]. A meta-analysis of seven trials also found lower depression scores [5].

Magnesium is not a conventional one-receptor drug. The biologically used species is Mg2+, which binds ATP and supports hundreds of enzymes involved in phosphorylation, ion transport, energy metabolism, protein synthesis and DNA synthesis [10]. That breadth explains why supplementation can change several systems when magnesium is low, but it also makes a single downstream mechanism for blood pressure, glucose or sleep unlikely.

Its clearest direct neural target is the NMDA receptor channel. At resting membrane potential, extracellular Mg2+ enters and blocks the open pore in a rapid, reversible and strongly voltage-dependent manner; depolarization relieves the block and permits Na+ and Ca2+ current [14]. This is normal synaptic gating, not evidence that an oral supplement acts like a high-affinity NMDA antagonist.

The voltage dependence is the informative detail. It is extreme, which places the magnesium binding site deep inside the channel pore rather than at a surface regulatory site, and it distinguishes magnesium's block from the slower, differently behaved block that dissociative anesthetics produce at the same receptor. Kainate and AMPA receptor channels, which pass the same sodium and potassium ions, carry no equivalent site [14]. Magnesium is not an NMDA drug; it is the reason the NMDA receptor acts as a coincidence detector, opening only when glutamate binding and depolarization arrive together.

Getting magnesium in and keeping it is a transport problem with named proteins. TRPM6, a transient receptor potential channel, handles active transcellular uptake in the intestine and reabsorption in the distal convoluted tubule; mutations in it cause primary hypomagnesemia with secondary hypocalcemia — a combined defect of gut absorption and renal conservation, which is how the gene was found. Bulk reabsorption happens elsewhere and by a different route: paracellin-1 carries magnesium and calcium between cells in the thick ascending limb of the loop of Henle, and the gamma subunit of the Na+/K+-ATPase turns out to be involved in the distal convoluted tubule [15]. Extracellular magnesium is set by intestinal absorption balanced against renal excretion, which is why kidney function and drugs acting on the tubule matter more to magnesium status than dose alone.

Magnesium's blood pressure effect tracks a small rise in serum magnesium. In one meta-analysis, 300 mg a day or one month of use was enough to raise serum levels and lower pressure [11]. Proposed downstream routes include calcium antagonism in vascular smooth muscle, endothelial function and electrolyte balance, but the clinical endpoint is modest and not formulation-specific.

Magnesium L-threonate is marketed for the brain. In a randomized trial in older adults with cognitive impairment, a formula based on it (MMFS-01) improved overall cognitive ability compared with placebo [16]. A 2025 trial of Magtein found better overall cognition and reaction time and a smaller improvement in self-reported sleep impairment. Ring-measured sleep did not change [17].

The mechanistic claim behind it comes from rats, not from a human head-to-head trial [18]. What that rat work reported is more specific than "more brain magnesium". Magnesium-L-threonate improved learning, working memory, and short- and long-term memory, and restored pattern-completion ability in aged animals. Hippocampal DG and CA1 showed a higher density of synaptophysin- and synaptobrevin-positive puncta, correlated with the memory improvement. Functionally, raised magnesium increased the number of functional presynaptic release sites while lowering each site's release probability — a reconfiguration that selectively strengthens transmission of burst inputs rather than single spikes. Together with upregulated NR2B-containing NMDA receptors and their downstream signaling, that enhanced the plasticity produced by correlated inputs [18].

Note the shape of that claim: the proposed benefit is a change in how synapses filter input patterns, not a general increase in excitation. It is also entirely preclinical.

Magnesium products differ in how much elemental magnesium they contain, how readily the salt dissolves, the amount that reaches blood or urine, and what the accompanying ion does. Those are separate questions; a higher urine value can mean more was absorbed and promptly excreted, not necessarily that more reached a target tissue [8]. A rat comparison found only modest salt-to-salt differences and cannot establish clinical superiority in people [19].

The defensible hierarchy is therefore modest: use the elemental amount, prefer a form you tolerate, favor soluble salts when absorption is the goal, and do not infer a clinical advantage from a short-term serum or urine curve alone [8].

Very large intakes can be dangerous. A 75-year-old woman died of hypermagnesemia after drinking milk mixed with magnesium chloride as a folk remedy [25].

In the trials cited here, magnesium was generally well tolerated, including the Magtein and bisglycinate trials [17][22]. Oral supplements commonly cause dose-limiting diarrhea [20]; impaired renal handling increases the risk that absorbed magnesium accumulates [10].

Other minerals taken in high doses, phytate, oxalate and some fibers reduce magnesium absorption. Protein and some non-digestible carbohydrates increase it [8].